Colors

The models are most readily compared to large samples of stars and brown dwarfs in color-magnitude diagrams. The standard system of broadband colors is sufficiently constraining when evaluating the accuracy of the models. This is because most spectral features are several thousands of angstroems wide, and the remaining emission windows are well sampled by each bandpass: Z at $1.0~\mu$m, J at $1.3~\mu$m, H at $1.6~\mu$m, K at $2.2~\mu$m, M at $4.5~\mu$m, and N at $10~\mu$m. Methane bands appear in brown dwarfs cooler than about 1700K at 1.7, 2.4 and 3.3 $\mu$m, reducing the flux sampled by the H, K and L' bandpasses respectively. The pressure-induced H₂ opacity, on the other hand, depresses the flux in the K bandpass in the coolest brown dwarfs and low-metallicity dwarf stars.

We have computed synthetic UBVRIJHKLL'M magnitudes by integrating our model spectra according to the photon count prescription at a wavelength step of 1Å. We have adopted the filter responses by and , bringing our synthetic photometry on the Cousins and Johnson-Glass system. Transformations to the CIT or to other systems are readily obtained from . As in previous papers, we used the energy distribution of Vega as observed by and to provide an absolute calibration. Zero magnitudes and colors are assumed for Vega. The grainless NextGen models of , as well as the AMES-Cond and AMES-Dusty models from this work are compared to the observed stellar and brown dwarfs samples of , , , , and in Figure . Please note that we have applied here a +0.18 dex shift in J-K to the current models to match the position of the NextGen models in the non-dusty regime in order to isolate the dust effects. This offset of the current models to the blue of the earlier NextGen models is due to some inaccuracies of the NASA-Ames H₂O opacity database in describing these relatively hot atmospheres (see Allard, Hauschildt and Schwenke 2000 for details).

These colors are interesting as they have helped distinguish interesting brown dwarf candidates from the databases of large scale surveys such as DENIS and 2MASS, and in obtaining an appreciation of the spectral sensitivity needed to detect new brown dwarfs. The methane bands cause the J-K colors of brown dwarfs to get progressively bluer with decreasing mass and as they cool over time. Yet their I-J colors remain very red which allows us to distinguish them from hotter low-mass stars, red shifted galaxies, red giant stars, and even from low metallicity brown dwarfs that are also blue due to pressure-induced H₂ opacities in the K bandpasses. Fortunately, grain formation and uncertainties in molecular opacities are far reduced under low metallicity conditions ([M/H]<-0.5). Therefore, model atmospheres of metal-poor subdwarf stars and halo brown dwarfs are free of uncertainties on the dust compared to their metal-rich counterparts. This has been nicely demonstrated by who reproduced closely the main sequences of globular clusters ranging in metallicities from [M/H]= -2.0 to -1.0, as well as the sequence of the halo subdwarfs in color-magnitude diagrams.

As can be seen from Figure , the AMES-Dusty models reproduce well the locus of the coolest dwarfs which deviate from that of main sequence stars red values of J-K as dust effects grow in their atmospheres with decreasing effective temperature (see also Leggett et al 1998 for more such color comparisons). It appears, therefore, that this full-dusty limit where grain settling is negligible is adequate to reproduce the global properties of late-type low-mass stars and young or massive brown dwarfs with ${\rm T}_{\rm eff}\le 1800$K. Below this temperature, the AMES-Dusty models keeps getting redder and do not correspond to the properties of known T dwarfs illustrated in this diagram by the position of Gl229B and SDSS1624.

The locus of the AMES-Cond models for their part depends upon two major uncertainties. The first, likely tied to the second, is a hump of flux excess between 0.8 and 0.93 $\mu$m, i.e. in the I-bandpass, which prevent the Cond models to become redder than I-J=4.2. The second is the description of the far wings of the absorption lines of K I and Na I D as discussed above and illustrated in Figure 9. In Figure 22 we show two grids of Cond models: one computed with a coverage of the line wings opacity contributions of 5000Å on each side of each atomic line core (long dashed line), the other computed with a maximum coverage of 15000Å (short dashed line). Both grids use Lorentz profiles for the atomic lines. Obviously, the profile of the optical Na I D and K I doublets is no longer Lorenzian beyond 5000Å from the line cores as also been noted found by . Since T dwarfs appear in the cone defined by these Cond models, an adequate theory of line broadening could be sufficient to reproduce their properties. Yet no theory exists for the treatment of the far wings of alkali elements broadened by collisions with H₂and helium species to this date . Until these become available, the present grid with a line wing coverage of 5000Å seem to provide an acceptable compromise and limiting case (with the Dusty models) for the spectroscopic properties of brown dwarfs.

. Until these become available, the present grid with a line wing coverage of 5000Å seem to provide an acceptable compromise and limiting case (with the Dusty models) for the spectroscopic properties of brown dwarfs.